A significant amount of research has suggested that disability is not a stable state over time. Individuals can move between states from no disability, to some or significant disability, or may show recovery over time. In a previous study, Medicare beneficiaries were categorized into disability levels based on self-reports of ADL and mobility limitations over a two year period. Categories included those who reported (1) no disability (self-report of no ADL or mobility limitation either year), (2) stability (some level of disability, but no change year to year), (3) increased disability, and (4) decreased disability. While these categories provided important initial descriptions of the sample, they did not provide a basis for understanding factors that contribute to changes in the level of disability or for identifying potential factors which could lead to effective interventions to impact disability level. In order to begin to understand the dynamic process of disablement, the costs associated with different transitions, the frequency of transitions (i.e. improvement or decline), and how limitations in functioning may be amenable to intervention, a scientific understanding of disability transition is needed. [unreadable] [unreadable] Transition probabilities have been calculated in a variety of samples to evaluate improvement after stroke, mobility difficulty with women with osteoarthritis, general health status, and even accumulation of deficits. Diehr and Patrick used transition probabilities to examine self-rated health states in a longitudinal study of older adults. Transition probabilities were found to vary by initial health state, age, and sex. Diehr and her colleagues (1998) also suggest that transitions can be used to estimate years of healthy life which can assist in determining cost-effectiveness of programs. More recent use of transitions has focused on disability in older ages, those 75 and older. However, the examination was limited to longitudinal data from 1982 to 1994 using states that included disabled, non-disabled, nursing home, and dead. Both nursing home placement and death were considered absorbing states in the model. There was no consistent pattern of trends, although results suggested declining rates of recovery. These results are in contrast to other research which has found lower prevalence of disability over time. This project is currently being initiated. The project timeline is in development, after which data analysis will commence.